INSIGHT is an autonomous AI agent designed for medical research, enabling users to automate literature reviews and data analysis. It targets researchers and power users seeking to streamline the process of gathering, synthesizing, and generating hypotheses from medical literature. The system offers a managed solution with more features at insightai.dev/project.

How It Works

INSIGHT employs a boss-worker agent architecture. A boss agent defines tasks based on an objective and previous results, storing them in a task list. Worker agents retrieve tasks, execute them using APIs like PubMed and MyGene, and leverage context from LlamaIndex for enhanced performance. Results are saved to LlamaIndex, informing the boss agent for iterative task refinement.

Quick Start & Requirements

• Install dependencies: pip install -r requirements.txt
• Set OPENAI_API_KEY environment variable or add to config file.
• Run: python main.py
• Requires Python and an OpenAI API key.

Highlighted Details

• Autonomous task creation, execution, and prioritization.
• Integrates with PubMed and MyGene APIs, with extensibility for more.
• Utilizes LlamaIndex for context and result storage.
• Supports reloading and continuing previous runs.
• Includes a talk_to_index.py for querying indexed data.
• Generates a key_findings.md with summaries, insights, and hypotheses.

Maintenance & Community

• Project initiated by oneil512.
• Open for contributions and inquiries via oneil512@umn.edu.

Licensing & Compatibility

• No license specified in the README.
• Compatibility for commercial use or closed-source linking is not detailed.

Limitations & Caveats

The project is presented as a personal project with an invitation for contributions, suggesting it may be in early development. Users should be mindful of potential costs associated with OpenAI API usage, especially with more powerful models.